svn commit: trunk/busybox/networking

[Bernhard Fischer]

On Sun, Jan 28, 2007 at 07:11:55PM +0100, Bernhard Fischer wrote:
>On Sun, Jan 28, 2007 at 06:59:12PM +0100, Denis Vlasenko wrote:
>>On Sunday 28 January 2007 18:59, Bernhard Fischer wrote:
>>> On Sun, Jan 28, 2007 at 06:25:38PM +0100, Denis Vlasenko wrote:
>>> 
>>> >Then why gcc doesn't warn me when I assign "const char*" to "char*"?
>>> 
>>> I don't understand?
>>
>>Try  char *str = "constant";
>
>$ cat foo.c 
>void f(void) {
>	char*c="constant";
>	char *str = c;
>	str[0]='\0';
>}
>$ gcc -c foo.c -Wwrite-strings
>foo.c: In function 'f':
>foo.c:2: warning: initialization discards qualifiers from pointer target type

mudflap doesn't catch this in runtime, of course (perhaps that's by design?),
but then, mudflap is a bit odd anyway:
Try to dump the help-text from a mudflapped app that has unresolved
symbols in one of it's dependent libraries some levels out of sight.

like: myapp
.         |
.          >liba.so
.                 |
.                  >libb.so
where libb.so has an unresolved sym.
You'll never ever be able to see
MUDFLAP_OPTIONS=-help ./myapp
since it aborts before it has parsed his (mudflaps) options.

I take it that this is a feature.. *shrug*